Distinction, Number, and the Empirical Filter: The Pre-Arithmetic Research Framework

Abstract

Several research lines in mathematical physics begin from identities between arithmetic

objects and partition functions, or between hierarchical structure and ultrametric geometry,

and ask whether any physical system realizes the resulting structure. Each such line must

answer the same two questions, and most failures in the literature come from answering them

carelessly: which of these structures is being claimed as real, and what observation would

show the claim false. This paper states the discipline once, as a reusable framework. It

constructs a nine-level ladder from the primitive of distinction to the empirical filter of

physics, with one construction operation between each pair of levels; it states two boundary

rules governing any movement up the ladder — a rule against uncommitted reification of the

primitive, and a rule that no mathematical isomorphism passes as a physical realization

without a stated measurement protocol, a null model, and a falsification condition; and it

defines a compact claim record that any research claim can carry, together with mechanical

demotion rules for claims that violate the boundary rules. The framework makes no empirical

claims of its own. It systematizes discipline that is currently distributed across seven

published records of the QNFO program [@umporr014; @res021; @res027; @res028; @res029; @res030; @res031], and it states the one claim it does make — that the

ladder covers the published lineage without remainder — together with a falsification

protocol, so the framework itself can be checked the way it asks everything else to be

checked.

1. Introduction and scope

The arithmetic-statistics program at QNFO has published, during 2026, a sequence of records

connecting number-theoretic structure to statistical mechanics. A gas whose modes are

indexed by primes, with logarithmic single-particle energies, has partition functions that

are exactly zeta objects: unrestricted occupation reproduces the Riemann zeta function, and

squarefree occupation reproduces the ratio of zeta functions at argument and double

argument [@res027]:

Here $Z{\mathrm{B}}$, $Z{\mathrm{F}}$, and $Z_{\mathrm{MB}}$ are the partition functions under

unrestricted, squarefree, and Boltzmann occupation respectively, and $P$ denotes the prime zeta

function. Bounded-occupation generalizations form a continuous family between the two

[@res028]. A companion record consolidated the correspondence and its practitioner-facing

reading [@res029], and a computational study then adjudicated whether the arithmetic cut can

be distinguished from non-arithmetic alternatives with matched level density [@res030].

Parallel work on hierarchy distance established that the number of distinctions required to

separate two objects is an ultrametric that does not depend on the realization — taxonomy,

p-adic digits, or Laurent coefficients — and that this distance is the canonical finite

distance of the program [@umporr014].

Each of these records had to solve, independently, the same methodological problem: where a

mathematical identity ends and a claim about the physical world begins. The present

framework is the extracted common discipline. It is a methodological scaffold, not an

empirical result: it makes no claim about what the world is like, and none of its nine

levels asserts anything ontic on its own.

The framework is motivated by a simple failure mode. Two extreme positions repeatedly

appear in work of this kind. One treats a combinatorial identity — a partition function

equal to a zeta function — as if it were already a statement about bosons or fermions. The

other treats all such identities as content-free. Both positions lose the same thing: the

controlled passage from a formal structure to an empirical claim, which is the only place

where the correspondence acquires content. The framework fixes what that passage requires,

in advance, for every claim in the program. A reader of any QNFO record can locate each of

its claims on the ladder, see the declared commitment level of each, and read exactly what

observation would refute it. This is the framework's purpose: to make the map-territory

boundary of an arithmetic-physics program legible, and its empirical commitments auditable,

without requiring every paper to rebuild the discipline from scratch.

2. The nine-level construction ladder

The framework organizes every research object by the level at which it is constructed. The

levels are ordered; each is built from the one below by a single named operation, and the

ordering is strict in the sense that skipping levels is not permitted without declaring

every intermediate step.

1. Distinction. A cut separating inside from outside. The minimal unit of structure;

the framework takes it as a primitive. Following the Laws of Form lineage

[@spencerbrown1969], a system is constituted by the distinctions that define it. Nothing

is asserted here about what the world is made of: the cut is a construction primitive of

the language, not a claimed constituent of reality.

1. Pre-arithmetic structure. Structure without number: order, hierarchy, partition,

adjacency. The distinction-count distance between two leaves of a rooted tree — the

number of cuts needed to separate them — is defined at this level: with $d(a,b)$ the

number of distinctions required to separate leaves $a$ and $b$, the strong triangle

inequality $d(a,c)\le\max\{d(a,b),\,d(b,c)\}$ holds, so the structure is an ultrametric

before any prime or valuation appears [@umporr014]. Ultrametric structure is

therefore pre-arithmetic: it exists whether or not counting exists.

1. Arithmetic. Counting and composition of distinctions. Concatenation of cuts gives

addition; iteration across independent cut-families gives multiplication; irreducible

multiplicative distinctions are the primes; unique factorization is the composition law.

The construction operation of this level is counting.

1. Number theory. Patterns of the composition: the distribution of irreducibles,

factorization statistics, and the generating functions of the composition, including

L-functions. The operation is pattern discernment on the output of counting.

1. Valuation. Assigning size to distinctions: norms and absolute values. By

Ostrowski's theorem there is one Archimedean place and one p-adic place per prime; the

p-adic valuation is one realization of hierarchy distance — not its ground, which lives

at level 2. The operation is sizing.

1. Geometry. The resulting relational form: metric and ultrametric spaces as the form

taken by valued distinctions, including their rigidity properties. The operation is

taking form.

1. Information. Distinction made operational: counting distinctions as bits, entropy,

and the localization of distinctions in two-point statistics rather than one-point

thermodynamic functions [@res030]. The operation is operationalization.

1. Measurement. The finite-resolution application of valuation to observation: which

observable, on which system, with which instrument, at which resolution and noise

budget — what a finite observer can actually distinguish. The operation is finite

resolution.

1. Physics. The empirical filter: falsification decides which of the structures are

real. The operation is filtering by observation.

The ladder is a translation device as much as a construction: each level corresponds to a

recognizable disciplinary home, which makes the framework legible outside its program of

origin.

Three structural rules complete the ladder. First, every claim, model, and interpretation

declares its level or its span of levels; a claim that cannot state its level is not yet a

claim. Second, work within a level is the default and needs no justification, while every

cross-level move requires a declared bridge. Third, movement downward is interpretation,

not derivation: a physical result may reinterpret an arithmetic object, but it does not

derive it. Violations of these rules — particularly conflating two objects that live at

different levels, such as prime-gap statistics and zero statistics [@res030] — are the

framework's primary defect class, and are adjudicated case by case in the records cited.

3. The two boundary rules

The ladder says where objects live; two boundary rules say what may be claimed about them.

3.1 Committed reification only

The primitive of distinction is methodological by default. Treating it — or anything built

on it — as a constituent of the world is a further commitment, and the framework requires

that commitment to be explicit, made per claim, made in advance, and accompanied by the

full realization requirements of Section 3.2. A commitment made after a null result has no

force; the discipline exists so that ontology is never smuggled into formalism

retroactively. The framework itself commits to nothing ontic anywhere on the ladder.

3.2 From isomorphism to realization

A mathematical isomorphism — however exact — is a map, not a territory. No claim may pass

from one to the other without three items stated in advance. First, a measurement

protocol: which observable, on which system, with which instrument, at which resolution.

Second, a null model: what the data would look like if the structure were absent. The

canonical pattern in this program is the matched-level-density null, in which synthetic

spectra carry the same smoothed level density as the target but none of its arithmetic

structure [@res030]. Third, a falsification condition: the observation, specified before

data collection, that would refute the claim. The framework states these three items as

one requirement because a protocol without a null model cannot say what a positive result

means, and a null model without a falsification condition cannot end an inquiry.

Two reporting rules accompany the requirement. Quantitative claims report effect sizes,

not only significance levels: a large deviation in one spectral window can carry little

information about the rest of the spectrum, and the reportable object is the full

discriminating curve, not a single threshold crossing [@res030]. And claims whose

observables are derived from one underlying two-point function — pair correlation, spectral

form factor, number variance — are one channel, not three independent confirmations, and

are corrected accordingly.

3.3 The claim record

Every claim that follows the framework carries eight fields: its ladder level; its carrier

(whether it is definitional, formal, computational, empirical, or engineered); its ontic

commitment (methodological by default, heuristic, or ontic); its map-territory status (map,

bridge, or territory); and, whenever the claim reaches for physical reality, the protocol,

null model, and falsification condition of Section 3.2. Claims about the framework itself

are permitted and carry the same record with level marked as meta-level.

The record is made auditable by mechanical demotion rules. Any bridge or territory claim

missing one of the three mandatory items — protocol, null model, or falsification

condition — is demoted one step per missing item, from territory to bridge to map, and no

further than map. An ontic commitment without the full triple is demoted to heuristic.

Where two violations apply to one claim, both demotions apply and the lower status wins.

The rules are total: every claim state has exactly one outcome and every demotion

terminates, which makes the record's integrity checkable by machine — a property the

deposited verification suite exercises exhaustively.

4. Relationship to the published records

The framework restates none of the results it systematizes. The hierarchy distance and its

realization independence are established in [@umporr014]. The exact partition-function

correspondence — unrestricted occupation giving the zeta function, squarefree occupation

giving its double-argument ratio — is established in [@res027] and extended to bounded

occupation in [@res028]. The consolidated map and its practitioner-facing reading are given

in [@res029]. The computational adjudication of the arithmetic cut against matched-density

nulls, including the location of the arithmetic information in two-point statistics, is

given in [@res030]. The companion consolidation record [@res031] audits the

dictionary and its five-level interpretive ladder at paper level; the present framework

generalizes that ladder to the nine levels above, which re-partition its upper half rather

than extending it, and the two records are kept consistent on that point.

The framework's level assignments to the published lineage are stated in the accompanying

source archive. Two assignments that the first version flagged as provisional — the measurement-level

reading of [@res029] and the span assigned to [@res021] — have been adjudicated: the

measurement-level reading was withdrawn in favor of an operational reading, and the span

is stated with its declared bridges. The adjudications, with per-record evidence,

accompany this version.

5. Verification

The framework's quantitative and mechanical content is checked by a deposited script that

(1) verifies the Euler-product identity underlying level 3 numerically, including the

squarefree ratio and the bounded-occupation endpoint; (2) verifies the ultrametric

triangle for the distinction-count distance on random hierarchies; (3) exhaustively checks

that the demotion rules of Section 3.3 terminate and assign exactly one outcome to every

claim state; and (4) checks the integrity of the ladder and of the cross-level assignment

table. The script is deterministic, dependency-free, and re-runnable from the deposited

layout; its full output is deposited alongside this paper.

6. The framework's own claim

The framework makes exactly one claim: that the nine-level ladder covers the published

lineage of the program without remainder — that every published object assigns to at least

one level or declared span. The claim fails if a published object cannot be assigned to

any level or span; it fails in a second, independent way if a defect already adjudicated in

the lineage turns out not to be adjudicable by the two boundary rules, which would show a

third rule type is needed. Both failure modes are specified in advance, the level

assignments are listed record by record in the source archive, and the provisional

assignments are named above. The framework asks of itself only what it asks of every claim

that passes through it.

References